Fractile CEO
LONDON, Nov. 11, 2026 | Fractile, a British startup building chips for running AI models, has raised a $720 million Series C round led by Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint, TechCrunch reported on Wednesday. The U.K. government put the company's valuation at "over £5 billion," which TechCrunch said could be about $7 billion, a figure reported by the Financial Times and a Reuters source.
Who Is Backing Fractile and What It Is Building
About a dozen other investors joined the round, including the U.K. Sovereign AI Fund, the British Business Bank and Alphabet's CapitalG, SiliconANGLE reported. The government said its fund's backing is aimed at "ensuring the U.K. has a greater stake in the hardware that will underpin its future economy," according to TechCrunch.
Fractile was founded in 2022 by AI researcher Walter Goodwin, its CEO. Its chips target inference, the work of running trained models in production. Large language models process prompts in many steps, and on today's hardware the results of each step travel to off-chip HBM memory and back, which slows things down. Fractile stacks standard DRAM directly on top of its processors so data has less distance to travel. Per reports cited by SiliconANGLE, the company expects up to 100 times better performance than current silicon and a 90% reduction in energy use. Fractile estimates its chips will let models process several thousand tokens per second.
The company is also building custom software for its systems and plans to ship its chips in rack-scale systems. Goodwin wrote on LinkedIn that "there is still a lot of work ahead" and that the funding will let Fractile "ship [its] first chips and systems next year," TechCrunch reported.
Where Fractile Fits in the AI Chip Race
Fractile joins Axelera, Etched and OLIX among the best-funded startups designing purpose-built AI chips, TechCrunch said. Others are attacking the same memory bottleneck in different ways. Cerebras Systems, which went public in May, keeps data in a large on-chip SRAM cache, and its newest WSE-3T processor holds 44 GB, according to SiliconANGLE. Samsung introduced zHBM in August, a memory design stacked on top of AI accelerators that it says will be eight times faster than HBM5. Our guide to Nvidia competitors covers the wider field.
For companies paying to run AI, the round matters more as a signal than as a product. Inference is the cost that grows with every query and every agent step, and TechCrunch noted that agentic AI, where models run longer chains of tasks, has made the speed and memory tradeoff more pressing. Investors and a national government are now betting heavily that a new chip design can lower that cost. The performance figures, though, are the company's own expectations for hardware that has not shipped, and no independent benchmarks were reported. Buyers will not be able to test Fractile systems before next year.
⭐ Get more AI Business Weekly in Google: add us as a preferred source on Google.
Frequently Asked Questions
How much did Fractile raise?
Fractile raised $720 million in a Series C round announced Nov. 11, 2026, led by Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint. The U.K. government put the valuation at over £5 billion, which TechCrunch said could be about $7 billion.
When will Fractile's AI chips ship?
Fractile expects to start shipping its first chips and rack-scale systems in 2027, according to CEO Walter Goodwin. The company says its design stacks DRAM directly on its processors to speed up AI inference and cut energy use.